Tony Arcieri

What are the next steps that need to happen towards stabilizing a feature like this? It would be very nice for [`crypto-bigint`](https://github.com/rustcrypto/crypto-bigint), where we have fairly aggressively leveraged `const fn`...

That'd be great! If there were an example to use as a starting point, I could probably help implement some of the ones we're interested in using. [_mulx_u32](https://doc.rust-lang.org/core/arch/x86_64/fn._mulx_u64.html) and [_mulx_u64](https://doc.rust-lang.org/core/arch/x86_64/fn._mulx_u64.html)...

After going down a very deep rabbit hole it seems ADX and MULX can't currently be emitted as intrinsics due to an LLVM bug: https://github.com/rust-lang/stdarch/issues/666 It seems the best path...

Regarding `const` and `asm!`, I'm curious if there might be some way to integrate `const_eval_select` into CPU feature detection macros like `is_x86_feature_detected!`. In the cases where we (prospectively) want to...

To fix this properly I think we need a SIGINT handler that shows the cursor again (and probably does a few other things like print a newline)

Here's how nio4r handles JRuby, if it's any help: https://github.com/celluloid/nio4r/blob/master/ext/nio4r/org/nio4r/Nio4r.java

@taichi something you might consider as a path to JRuby support is rewriting the core of cool.io to use nio4r. I would be happy to add features upstream to nio4r...

The MAC can be appended or prepended to the ciphertext. It would not be possible for an attacker to compute a valid MAC without knowing the key.

If you compute the HMAC of the plaintext, you're doing encrypt-and-MAC. This was used by SSH, and is vulnerable to chosen ciphertext attacks. See e.g.: http://eprint.iacr.org/2002/078.pdf What you'd want to...

If you're doing RSA encryption exclusively, you should be fine with OAEP. A MAC is needed for symmetric encryption.